Page 4 of 6 [ 84 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6  Next

n4mwd
Veteran
Veteran

User avatar

Joined: 7 Jun 2008
Age: 57
Gender: Male
Posts: 1,244
Location: Palm Beach, FL

06 Sep 2008, 10:17 am

Fuzzy wrote:
No I think you took that the opposite way.

from http://www.tomshardware.com/forum/240282-28-excited
Quote:
Flash wear out is not a concern with SSD's. Someone at StorageSearch.com crunched the numbers and determined that a 160GB SSD drive with an 80MB/s throughput run at 100% write utilization would take 51 years exhaust the write limit of the flash memory. That's a brutal worst-case scenario. In real life, reads and writes average about 3:1, which means the predicted failure will take over 80 years.


so I looked it up at storageseach.com. This is what I found.

http://www.storagesearch.com/ssdmyths-endurance.html


Don't believe it. They are using spin calculations. As in, they are spinning truth out of lies.

What they are saying is if you write values to every location in an entire drive, then it will take 51 years before flash burnout. This creates the illusion that the drive will last that long. Notice how they use the size of the drive in the calculation. The size of the drive is irrelevant. It only takes one bad flash location to ruin the entire drive.

For example, suppose you have a clock application that saves the current time to a file on the disk every second. This means one write every second. Since the flash devices burn out after 2 million writes to a memory CELL - not 2 million writes to the entire drive - the burnout can be calculated as follows:

2M / 86400 = 23 days. Where 86400 is the number of seconds in a day.

Modern operating systems generally use a lot of paged memory (swap partition or swap file). The eee surf doesn't seem to use that, but most linuxes do and so does XP. This virtual memory swapping does a lot of disk writing.

What that means is that the ssd would probably fail if used as a server without an external hdd.

With that said, I got mine and it seems to run fine for now. Its underclocked and I need to figure out how to set it back to its 900 mhz advertised rate. It came with instructions on how to get it to run with XP. Unfortunately, the driver CD for XP was in spanish so I have to call the company and get an English one.



n4mwd
Veteran
Veteran

User avatar

Joined: 7 Jun 2008
Age: 57
Gender: Male
Posts: 1,244
Location: Palm Beach, FL

06 Sep 2008, 10:23 am

Fuzzy wrote:
Maybe I should make a club for cute little computers?

But you are the only one that has one yet!


I have one too. I got mine at buy.com but you can literally buy these things at Toys R Us.

I'm not convinced about the longevity of the SSD though.



MR_BOGAN
Veteran
Veteran

User avatar

Joined: 5 Mar 2008
Age: 119
Gender: Male
Posts: 3,200
Location: The great trailer park in the sky!

06 Sep 2008, 6:31 pm

n4mwd wrote:
Fuzzy wrote:
Maybe I should make a club for cute little computers?

But you are the only one that has one yet!


I have one too. I got mine at buy.com but you can literally buy these things at Toys R Us.

I'm not convinced about the longevity of the SSD though.


Welcome to the club. :thumright:

Our cute computer club is currently been infultrated so we are going to have to put you through an inititation process to make sure you are not a spy, and that you are NOT ONE OF THEM!! !!



computerlove
Veteran
Veteran

User avatar

Joined: 10 Jul 2006
Age: 119
Gender: Male
Posts: 6,701

06 Sep 2008, 6:48 pm

speaking of initiation, years ago I saw the elastic band of a brief (underwear) while walking the uni campus ... 8O


_________________
One of God's own prototypes. Some kind of high powered mutant never even considered for mass production. Too weird to live, and too rare to die.


lau
Veteran
Veteran

User avatar

Joined: 17 Jun 2006
Age: 71
Gender: Male
Posts: 10,595
Location: Somerset UK

06 Sep 2008, 7:22 pm

n4mwd wrote:
...What they are saying is if you write values to every location in an entire drive, then it will take 51 years before flash burnout....

This is correct.

The technique used is that the flash memory is recycled progressively. I.e. as new writes occur, the area written "walks" around the memory.

If the drive is nearly full of data, with only a small free area being repeatedly reused, this will reduce the overall lifetime. I.e. with 90% in use and unchanging, the 51 year estimate would reduce to 5.1 years.

So... strangely counter-intuitive as it may sound, your flash drive will last LONGER, the more often you overwrite it (as in, ALL of it).


_________________
"Striking up conversations with strangers is an autistic person's version of extreme sports." Kamran Nazeer


Fuzzy
Veteran
Veteran

User avatar

Joined: 30 Mar 2006
Age: 47
Gender: Male
Posts: 5,532
Location: Alberta Canada

07 Sep 2008, 12:15 am

Agreed. And if a segment goes bad, you should just get a bad sector like in a regular hard drive. That will be marked as off limits to read/writes, rather than being a catastrophic failure.


_________________
davidred wrote...
I installed Ubuntu once and it completely destroyed my paying relationship with Microsoft.


Fuzzy
Veteran
Veteran

User avatar

Joined: 30 Mar 2006
Age: 47
Gender: Male
Posts: 5,532
Location: Alberta Canada

07 Sep 2008, 2:21 am

n4mwd wrote:

Modern operating systems generally use a lot of paged memory (swap partition or swap file). The eee surf doesn't seem to use that, but most linuxes do and so does XP. This virtual memory swapping does a lot of disk writing.


I respectfully disagree with you.

Swaps are becoming smaller in comparison to overall ram size. And A lot of people dont use them at all. This xp machine, with 3 gigs of ram ram, works faster without a page file. My linux machines, with 1, 1.5 and 2 gigs or ram respectively, never seem to access their swap partitions at all. I wasnt even aware that my 2 gig machines swap wasnt mounted for for the first year.


_________________
davidred wrote...
I installed Ubuntu once and it completely destroyed my paying relationship with Microsoft.


Enigmatic_Oddity
Veteran
Veteran

User avatar

Joined: 4 Nov 2005
Age: 33
Gender: Male
Posts: 2,604

07 Sep 2008, 2:56 am

I was under the impression that Windows XP couldn't run without a page file, regardless of the amount of installed RAM.



Fuzzy
Veteran
Veteran

User avatar

Joined: 30 Mar 2006
Age: 47
Gender: Male
Posts: 5,532
Location: Alberta Canada

07 Sep 2008, 3:40 am

Enigmatic_Oddity wrote:
I was under the impression that Windows XP couldn't run without a page file, regardless of the amount of installed RAM.


Yes you can. This page sums it up.
"If I have a Windows XP machine that has lots of memory, can I improve performance by removing the pagefile?"
I happen to disagree with it in regards to performance.


_________________
davidred wrote...
I installed Ubuntu once and it completely destroyed my paying relationship with Microsoft.


n4mwd
Veteran
Veteran

User avatar

Joined: 7 Jun 2008
Age: 57
Gender: Male
Posts: 1,244
Location: Palm Beach, FL

07 Sep 2008, 7:35 am

lau wrote:
n4mwd wrote:
...What they are saying is if you write values to every location in an entire drive, then it will take 51 years before flash burnout....

This is correct.

The technique used is that the flash memory is recycled progressively. I.e. as new writes occur, the area written "walks" around the memory.

If the drive is nearly full of data, with only a small free area being repeatedly reused, this will reduce the overall lifetime. I.e. with 90% in use and unchanging, the 51 year estimate would reduce to 5.1 years.

So... strangely counter-intuitive as it may sound, your flash drive will last LONGER, the more often you overwrite it (as in, ALL of it).


What you are describing is an algorithm to extend the life of a write limited device. This indeed does extend the life of the device as a whole, but as you mentioned, its not that effective if the device is at capacity. Further life could be realized if used sectors that don't get rewritten very often are swapped with sectors that are nearing failure. The list of techniques goes on and on, but all they do is extend the life of the device a little bit.

What you say would indeed work, but I have not seen any evidence that these chips actually work that way. A link would be appreciated as there isn't much on the factory website. Its also not clear exactly who is supposed to be doing all of that - the OS or the chip.

From The Website wrote:
We assume perfect wear leveling which means we need to fill the disk 2 million times to get to the write endurance limit. [Emphasis added.]

2 million (write endurance) x 64G (capacity) divided by 80M bytes / sec gives the endurance limited life in seconds.

That's a meaningless number - which needs to be divided by seconds in an hour, hours in a day etc etc to give...

The end result is 51 years!


Note that their example assumes that all you do is continuously overwrite the entire device over and over. Again, note that they are talking about a 64GB device which takes 800 seconds to fully write. However, the Surf does not have 64GB, it has 2GB. It only takes 25 seconds to fully write.

So using their very optimistic best case scenario formula, 2M x 2G / 80M = 1.5 years!! !

Coincidentally, the warranty is only one year.



Fuzzy
Veteran
Veteran

User avatar

Joined: 30 Mar 2006
Age: 47
Gender: Male
Posts: 5,532
Location: Alberta Canada

07 Sep 2008, 8:06 am

n4mwd wrote:
So using their very optimistic best case scenario formula, 2M x 2G / 80M = 1.5 years!! !

Coincidentally, the warranty is only one year.


VERY good point.


_________________
davidred wrote...
I installed Ubuntu once and it completely destroyed my paying relationship with Microsoft.


Vexcalibur
Veteran
Veteran

User avatar

Joined: 17 Jan 2008
Age: 34
Gender: Male
Posts: 5,573

07 Sep 2008, 10:02 am

1.5 years of continuous I/O.

Every computer deserves to rest after that. Since these have USB ports you probably can buy a new disk from which to run the OS, this is just a theory though, I have no idea.


_________________
.


lau
Veteran
Veteran

User avatar

Joined: 17 Jun 2006
Age: 71
Gender: Male
Posts: 10,595
Location: Somerset UK

07 Sep 2008, 11:17 am

n4mwd wrote:
...
What you say would indeed work, but I have not seen any evidence that these chips actually work that way. A link would be appreciated as there isn't much on the factory website. Its also not clear exactly who is supposed to be doing all of that - the OS or the chip...

I'm not sure what is done at present. Bear in mind that once upon a time, hard drives had no code inside them, whereas now, they are quite smart.

It certainly is no business of the OS, per se, or even the higher levels of the device driver (which are trying to make the chip appear like a "normal" HDD, in effect).

I would expect the cell management to be implemented at the lowest, hardware specific, layer of the driver. (If not in the chip itself.)

=======
On a bit of investigation, it would appear that the chips indeed do do their own management, these days:
http://en.wikipedia.org/wiki/Wear_levelling
http://wiki.eeeuser.com/ssd_write_limit
=======

I was peripherally involved in a project a decade ago: http://tfs.firshman.co.uk/ql/romdisq.htm

We did our sums... and decided that under typical usage, the limits were not a problem.

The driver did indeed do exactly as I described... in point of fact, I seem to recall that it kept an exact count of the number of writes to every cell.

==========

All the above certainly doesn't mean that you can't come up with a scenario that will burn out ONE CELL of a chip out in a minimal time, but that doesn't really matter.


_________________
"Striking up conversations with strangers is an autistic person's version of extreme sports." Kamran Nazeer


n4mwd
Veteran
Veteran

User avatar

Joined: 7 Jun 2008
Age: 57
Gender: Male
Posts: 1,244
Location: Palm Beach, FL

08 Sep 2008, 8:12 pm

I went to the eeuser site and I have some problems with his figures. He is claiming that the flash cells fail at 100K-200K and not 1M - 2M. Most MFG's claim the latter. The former is typical of chips made 20 years ago. Also, he is claiming that the max write speed that the eee pc is only 2-3MB/s. I admit I haven't run tests myself, but I just downloaded EEEBuntu for it and it downloaded at about 132 MB/s on another machine with a slower processor. So I'd say that 2-3MB/s seems a little low.



Vexcalibur
Veteran
Veteran

User avatar

Joined: 17 Jan 2008
Age: 34
Gender: Male
Posts: 5,573

11 Sep 2008, 4:52 pm

I got one! A 4G Asus 701, it is actually the first time I actually have a webcam.

It looks good so far, btw, regarding the SSD life, You can actually boot an operating system from an external USB flash disk! I also think you can replace the SSD, not sure though - Just saying: if the SSD died it would still not be the end of the computer's life.


_________________
.


Last edited by Vexcalibur on 11 Sep 2008, 11:37 pm, edited 1 time in total.